1 / 12

Teaching Survey Sampling Theory using R

This poster session explores the effective use of R in teaching Survey Sampling Theory at George Washington University. It highlights various applications such as data analysis, exploration, programming complex formulas, and simulation of estimators' properties. Key tools like boxplots, histograms, and numerical summaries are discussed for data examination. The importance of understanding bias, variance, and mean skewness in simulation results is emphasized. Suggestions for improving course delivery include consistent syntax, clearer variable definitions, and practical project-based learning.

heba
Télécharger la présentation

Teaching Survey Sampling Theory using R

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript

  1. Teaching Survey Sampling Theory using R Michael D. Larsen George Washington University UseR 2010 poster session, 7/21/10

  2. Uses of R in the course • Data analysis; exploring data • Programming complex formulas • Simulation of properties of estimators • Make estimation easier so one can think about concepts

  3. Exploring data • Examine means, sizes of clusters: more variability increases variance • Examine means, sizes across strata: more variability decreases variance • Examine skewness of variables: extreme skewness in population can lead to unrealistic sample-based estimates

  4. Exploring Data: Tools • Side-by-side boxplots • boxplot(split(senic$nurses, senic[,c("region","medical")]), xlab="four regions in U.S.; two hospital types", ylab="# nurses", main="113 hospitals in U.S.") • Histograms • Numerical summaries • Correlations; regression • sapply for lists created using ‘split’ command

  5. Comparing two factors for stratification potential

  6. Programming complex formulas • Checks understanding of formulas • Helps memorization of formulas • Next page: two-stage cluster sample estimator for total and variance of total

  7. Simulation • Using functions, one can contrast complex estimation methods in terms of bias, variance and MSE • Simulating 1,000 samples and plotting results gives different impression than mathematical result; Impact of skewness and outliers is more transparent

  8. Ease of use • Make estimation easier so one can think about concepts; Possible to focus on contrasts and more variables • Students can do more ambitious projects and handle ‘real’ data

  9. Ease of use example For a given budget and population, what is the advantage of more clusters with smaller sample sizes versus fewer clusters with bigger sample sizes? • Compute variances under three scenarios. • Take 10,000 samples under three different scenarios and compute variance of estimates. • Apply to three different variables. Write a summary.

  10. Suggestions, part 1 • Consistent syntax across survey designs • Ease of use: be clear on what time of variables are needed – factor, numeric, etc. • More examples with more numbers that can be replicated

  11. Suggestions, part 2 • Recover formula – not only R syntax but also an estimation formula – when run command • More details in context when errors occur: • Your sample sizes for clusters (ni) exceed your population sizes for clusters (Ni). • Only one primary sampling unit (defined by psu) is available for some clusters • Include writing projects based on data analysis

More Related